no-cap bass fix
joerg at openmoko.org
Wed Mar 24 16:22:15 CET 2010
This is *STRONGLY DISCOURAGED* and will break quite a couple of things.
Details see inline below
[Stefano Cavallari Di 23. März 2010]:
> Yesterday night I was going to fix the poor audio response of the
> Just before starting to solder (having opened the phone and the metallic
> plate) I discovered the caps I got were the wrong ones.
> So I looked at the scheme for an alternative solution, and I decided to try
> replace the audio caps with 0R, thus losing DC blocking.
You're not only losing the DC-decoupling, you're also losing the negative half
of the sine wave, when amp is basically shorting output to GND to create the
negative current by discharging the coupling capacitor. See operation
principles of analog bridge amp outputs in some good book about electronics.
This is NOT a class-A amp
(http://en.wikipedia.org/wiki/Electronic_amplifier#Class_A ), means it never
opens both the pullup and pulldown transistor concurrently. The LM4853 is a
class-B push-pull bridge.
> The plan was to measure the DC component and if low enough, leave it as it.
> If not, putting the DC filter in the minijack adapter.
This is basically feasible, but will most surely break JACK_INSERT logic, by
applying a voltage >0V to the GPIO detecting if a jack is inserted or not.
Each time you enable the amp to output some sound the headphones, it will
latch up and not detect jack removal. And it's quite unlikely jack insertion
is correctly detected each time as well, there also might be both false
positives and false negatives.
> So I did that, and it seem to work. I tried first with a multimeter. It
> 0.2 V DC, but I have to confirm it with an oscilloscope.
This reading probably is with headset amp disabled. Correct reading should be
Vmid, i.e. ~1.6V
> I tried the audio with very cheap headphones first (I was afraid of burning
Chances are you actually will end up with broken headset speakers, just
because of this
> then with decent ones. It seems to work way better!
> Now I just need a better adapter cable (mine need to be inserted middle way,
> it's not the right one), and then I have usable audio :)
Don't you think, if this was a viable workaround for the problem, we at OM
(particularly me in this case) came up with this suggestion some 1.5 .. 2
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 198 bytes
Desc: This is a digitally signed message part.
Url : http://lists.openmoko.org/pipermail/community/attachments/20100324/43185b63/attachment-0001.pgp
More information about the community